Self-tuning of Data Allocation and Storage Management: Advantages and Implications

Many important business applications use complex database management systems (DBMS). These DBMS have to be administrated and optimized for an optimal performance, especially in time-critical applications. Administration and optimization are very complex and costly tasks. Therefore, researchers and DBMS vendors focus on development of self-tuning techniques for a continuous adaption, e.g., the COMFORT automatic tuning project [6] or the MAPE approach by IBM [2]. However, the optimization and usage of self-tuning techniques for allocation and storage management of data are less investigated. DBMS vendors inform their customers about advantages and disadvantages of optimizing data storage in their user manuals and tuning guidelines. They recommend usage of functionality to optimize data storage with respect to the higher administration costs. But, DBMS vendors do not publish guidelines for this. Optimization of data storage is a complex (high administrative needs) task with many of options and parameters. For instance, the number of parameters for table space creation is huge, e.g., page size or database partition group. These two parameters affect essentially the performance of data allocation and storage management. In the scope of data allocation and storage management, many parameters dependencies and implications are unobserved. Our approach will observe the affect of the parameters. This paper presents first steps for better solutions and estimations for data allocation and storage management based on the parameter page size and its configuration.